Manuel Baltieri @manuelbaltieri.bsky.social · 19/09/2026Happy to see this coming together. Great work by Anil, pushing his ideas out there to foster a much needed discussion. I remain unconvinced by any of Anil's arguments in favour of "biological naturalism", and yet see no good evidence at the moment for machine consciousness. 1/ 100
Manuel Baltieri @manuelbaltieri.bsky.social · 13/07/2026This time, celebrating 19 days of being “on hold” on arXiv! “A Braitenberg vehicle’s habitat”, w/ @frosas.bsky.social and Filippo Torresan. Accepted at ALIFE 2026. What part of the environment constitutes an agent’s effective world, its habitat? 1/ 181
Manuel Baltieri @manuelbaltieri.bsky.social · 10/07/2026New work led by Filippo out on Neurocomputing! W/ @ksks.bsky.social , @kanair.bsky.social and me. Does an agent need to be "aware" of its executed actions to learn to achieve a certain goal? In reinforcement learning, the answer is generally "yes". 1/ 151
Manuel Baltieri @manuelbaltieri.bsky.social · 02/07/2026New preprint w/ @nathanielvirgo.bsky.social: Bayesian updates from coalgebraic determinisation. We show a relation between Bayesian filtering updates and some classical automata theory constructions. The results are a bit technical, so I'll focus on some applications below. 1/ 270
Manuel Baltieri @manuelbaltieri.bsky.social · 29/06/2026Celebrating 36 days of being kept "on hold" on arXiv! Paper: manuelbaltieri.com/assets/pdf/W... World models are a pretty hot topic right now, but what is a world model? Joint work w/ Filippo, @yivannaviy, Alec, and @_fernando_rosas 1/ 131
Manuel Baltieri @manuelbaltieri.bsky.social · 26/06/2026Our work on compositional behavioural semantics for state abstraction was accepted at ICML! Link: arxiv.org/abs/2606.25357 State abstraction compresses an RL state space while trying to retain decision-relevant behaviour. What survives? 1/arxiv.orgCompositional Behavioral Semantics for State Abstraction in Reinforcement LearningState abstraction plays a key role in scaling reinforcement learning to complex but structured systems. In studying such systems, a wide range of behavioral structures have been studied in reinforceme... 1102
Reposted by Manuel BaltieriMariel Goddu @marielgoddu.bsky.social · 11/07/2025My comment on Fillipo Torresan & @manuelbaltieri.bsky.social's "Disentangled representations for causal cognition" in Physics of Life Reviews: www.sciencedirect.com/science/arti... I argue that there is little meaningful analogy between learning from "pixels" vs "experience," but I praisesciencedirect.comDisanalogies between causal learning in animals vs. machines: Comment on “disentangled representations for causal cognition” by F. Torresan & M. BaltieriNone. 163
Reposted by Manuel BaltieritheHigherGeometer @highergeometer.mathstodon.xyz.ap.brid.gy · 02/05/2025Igor, you legend. Don't stop being you. There are ten more of these unreadable hypercube diagrams on the following pages.... Souce: arxiv.org/abs/2505.00682 114
Reposted by Manuel BaltieriAnna Abalkina @abalkina.bsky.social · 25/04/2025My experience applying for retractions at Elsevier. I've looked at paper mills since 2019 and drafted a preprint on a paper mill from an international publisher in 2021. I started contacting journals or research integrity teams to raise concerns about papers. Publishers react differently. 1/n 23116
Reposted by Manuel BaltieriJohannes Kleiner @johanneskleiner.bsky.social · 23/04/2025Great talk by @manuelbaltieri.bsky.social! 021
Reposted by Manuel BaltieriGordon @gordon.bsky.social · 10/04/2025"We discuss the problem of running today’s software decades, centuries, or even millennia into the future" tinlizzie.org/VPRIPapers/t...tinlizzie.org 052
Reposted by Manuel BaltieriFernando Rosas @frosas.bsky.social · 08/04/2025Preprint time: “AI in a vat: Fundamental limits of efficient world modelling for agent sandboxing and interpretability” arxiv.org/abs/2504.04608 Exploring the fundamental limits that shape the design space of world modelling for agent sandboxing and interpretabilityarxiv.orgAI in a vat: Fundamental limits of efficient world modelling for agent sandboxing and interpretabilityRecent work proposes using world models to generate controlled virtual environments in which AI agents can be tested before deployment to ensure their reliability and safety. However, accurate world m... 1146
Reposted by Manuel BaltieriJosé A. Alonso @jalonso.eurosky.social · 02/04/2025GOL in GOL in HOL: Verified circuits in Conway's game of life. ~ Magnus O. Myreen, Mario Carneiro. arxiv.org/abs/2504.00263 #ITP #HOL4arxiv.orgGOL in GOL in HOL: Verified Circuits in Conway's Game of LifeConway's Game of Life (GOL) is a cellular automaton that has captured the interest of hobbyists and mathematicians alike for more than 50 years. The Game of Life is Turing complete, and people have be... 031
Manuel Baltieri @manuelbaltieri.bsky.social · 26/03/2025Great to see a colleague speaking up, sad to think about the state of affairs. 010
Reposted by Manuel BaltieriᛕᎥᕼᗷᗴᖇᑎᗴ丅Ꭵᑕᔕ @kihbernetics.bsky.social · 23/03/2025I don't want to delete anything. I simply agree with Barbieri's distinction and claim that for a successful syntactic relationship, there is no need for anticipation or computation. On that level, the cell is a simple reliable #state machine (transducer) with no place for interpretation of meaning. 121
Reposted by Manuel BaltieriComplexity Cat 🐱 @amahury.bsky.social · 14/03/2025One of the most controversial corollaries of relational biology is the impossibility of simulating life. But what if I tell you that this claim is simply the result of misinterpreting Robert Rosen's ideas? #complexitycat 🐈⬛👇🧵1/3 amahury.github.io/posts/trilog...amahury.github.ioRelational Biology I: Is it possible to simulate life?The first part of this trilogy is devoted to discussing the difference between model and simulation, one of the cornerstones for understanding relational biology. How true is it that Robert Rosen deni... 2105
Reposted by Manuel BaltieriKonrad Kording @kordinglab.bsky.social · 12/03/2025Re the Tononi paper: Both Tononi’s IIT (phi) and Friston’s FEP start from fundamental, axiomatic, and debatable assumptions. These assumptions are generally made without any humility. This logic allows them to make exceptionally broad claims. Which contributes to my unease about them. 99117
Manuel Baltieri @manuelbaltieri.bsky.social · 10/03/2025Directed wiring diagrams for Mealy machines! 010
Reposted by Manuel Baltierilastpositivist.bsky.social @lastpositivist.bsky.social · 08/03/2025Only just learning about this now -- I guess for a while people have predicted that the AI doomer rationalist crowd would go violent, so its not surprising in some sense. Still though, odd times! www.theguardian.com/global/ng-in...theguardian.comThey wanted to save us from a dark AI future. Then six people were killedHow a group of Silicon Valley math prodigies, AI researchers and internet burnouts descended into an alleged violent cult 109115
Manuel Baltieri @manuelbaltieri.bsky.social · 05/03/2025After a multi-year collaboration with Martin Biehl, @mattecapu.bsky.social and @nathanielvirgo.bsky.social, I’m stoked to share the first of (hopefully) many outputs: “A Bayesian Interpretation of the Internal Model Principle” arxiv.org/abs/2503.00511. 1/arxiv.orgA Bayesian Interpretation of the Internal Model PrincipleThe internal model principle, originally proposed in the theory of control of linear systems, nowadays represents a more general class of results in control theory and cybernetics. The central claim o... 1196
Manuel Baltieri @manuelbaltieri.bsky.social · 25/02/2025This looks fantastic, and goes to the must read pile for 2025 081
Reposted by Manuel BaltieriDavid Gilbert @davidgilbert.bsky.social · 17/02/2025Elon Musk is making wild claims that #DOGE has found evidence of massive Social Security fraud.... The reality is that Musk's young engineers don't seem to be able to understand COBOL databases while Musk is misrepresenting the Social Security data www.wired.com/story/elon-m...wired.comNo, 150-Year-Olds Aren't Collecting Social Security BenefitsElon Musk claims to have found rampant fraud in the Social Security Administration. There's a much simpler explanation. 12727421013
Manuel Baltieri @manuelbaltieri.bsky.social · 17/02/2025More on process theories arxiv.org/abs/2502.10368arxiv.orgGeneralised Process TheoriesProcess theories provide a powerful framework for describing compositional structures across diverse fields, from quantum mechanics to computational linguistics. Traditionally, they have been formaliz... 170
Manuel Baltieri @manuelbaltieri.bsky.social · 09/02/2025Generalising Laplacians in a categorical setup 040
Manuel Baltieri @manuelbaltieri.bsky.social · 08/02/2025Too many things to read already for this year arxiv.org/abs/2404.14423arxiv.orgA Mereological Approach to Higher-Order Structure in Complex Systems: from Macro to Micro with MöbiusRelating microscopic interactions to macroscopic observables is a central challenge in the study of complex systems. Addressing this question requires understanding both pairwise and higher-order inte... 181
Reposted by Manuel BaltieriJosé A. Alonso @jalonso.eurosky.social · 08/02/2025Review of "Haskell in depth" by Vitaly Bragilevsky. ~ Brent Yorgey. www.cambridge.org/core/service... #Haskell #FunctionalProgramming 072
Reposted by Manuel BaltieriGabriel Peyré @gabrielpeyre.bsky.social · 29/01/2025This review paper by @guillaume-garrigos.com on SGD-related algorithms is a fantastic resource, offering elegant, self-contained, and concise proofs in a single, accessible reference. arxiv.org/pdf/2301.11235 118940
Manuel Baltieri @manuelbaltieri.bsky.social · 07/02/20252025 already clearly not disappointing for categorical systems theory arxiv.org/abs/2410.10627arxiv.orgEffectful Mealy Machines: Bisimulation and TraceWe introduce effectful Mealy machines - a general notion of Mealy machine with global effects - and give them semantics in terms of both bisimilarity and traces. Bisimilarity of effectful Mealy machin... 020
Reposted by Manuel BaltieriRodney Brooks @rodneyabrooks.bsky.social · 06/02/2025Oh gosh. Found on the web. Anthropic (producer of Claude LLMs) asking job applicants to not use AI systems when filling in their application. simonwillison.net/2025/Feb/2/a...simonwillison.netA quote from AnthropicWhile we encourage people to use AI systems during their role to help them work faster and more effectively, please do not use AI assistants during the application process. We … 1428
Reposted by Manuel BaltieriTechCrunch @techcrunch.com · 19/01/2025AI benchmarking organization criticized for waiting to disclose funding from OpenAItcrn.chAI benchmarking organization criticized for waiting to disclose funding from OpenAIAn organization developing math benchmarks for AI didn’t disclose that it had received funding from OpenAI until relatively recently, drawing allegations of impropriety from some in the AI community. Epoch AI, a nonprofit primarily funded by Open… 2939
Reposted by Manuel BaltieriPablo Samuel Castro @pcastr.bsky.social · 13/01/2025What are some known/published *active* applications of RL in production (i.e. are actually providing some financial utility)? Note that this excludes games, simulations, and in-lab robotics. 11428
Reposted by Manuel BaltieriALIFE Conference 2025 @alife2025.bsky.social · 05/01/2025Save the date! The annual #Conference on #ArtificialLife - #ALIFE2025 will take place in the heart of Kyoto, Japan, from 6-10 October 2025. The conference logo, official website launch, and more details coming soon! 16230
Reposted by Manuel BaltieriClément Canonne @ccanonne.github.io · 10/01/2025💡 Making slides, giving talks, and looking for illustrations for Alice, Bob, and the whole crew? Simón Oya (UBC) has drawn fresh, beautiful ones and releasing them under CC BY-SA 4.0 license: simonoya.com/drawings/ (discovered through the #EnCOREPrivacy25 workshop) 0227
Reposted by Manuel BaltieriOCaml @ocaml.org · 08/01/2025dlvr.itThe Most Elegant Configuration LanguageCCL: Categorical Configuration Language 082
Manuel Baltieri @manuelbaltieri.bsky.social · 06/01/2025Saw a discussion on metaphor-based metaheuristics. It's funny how I initially got into #AI / #ALife, through bio-inspired #optimisation, and now it turns out that the entire field has barely moved for decades, and most of the "results" in the last two decades are part of various running jokes. 1/ 282
Manuel Baltieri @manuelbaltieri.bsky.social · 05/01/2025Italians and their (our) mozzarella phase arxiv.org/abs/2501.005...arxiv.orgPhase behavior of Cacio and Pepe sauce"Pasta alla Cacio e pepe" is a traditional Italian dish made with pasta, pecorino cheese, and pepper. Despite its simple ingredient list, achieving the perfect texture and creaminess of the sauce can ... 051
Manuel Baltieri @manuelbaltieri.bsky.social · 04/01/2025Gemini remains an absolute mystery to me: so far behind all its competitors, and it’s not even close. (I’m not using any other LLM for work, but I can say other models are at least fun to play with here and there, when I got access to Gemini on Google workspace, it was just rubbish.) 010
Reposted by Manuel Baltieriitsthemind.bsky.social @itsthemind.bsky.social · 25/12/2024Ants Solving A Puzzle Credit: "Comparing cooperative geometric puzzle solving in ants versus humans" (scientific paper, doi.org/10.1073/pnas..., Tabea Dreyer, Amir Haluts Amos Korman and Ofer Feinerman) When ants work in groups, their performances rise significantly. 23212
Reposted by Manuel BaltieriJencel Panic @abuseofnotation.mathstodon.xyz.ap.brid.gy · 24/12/2024#categorytheoryillustrated got a design makeover, plus there is some new content, at the ending of the second and fifth chapters. I also started distributing a epub and pdf version of my books which are available for my Ko-fi and Patreon supporters […] [Original post on mathstodon.xyz] 23713
Reposted by Manuel BaltieriSanta Fe Institute @sfiscience.bsky.social · 23/12/2024Knowing only the building blocks of our own biosphere, can we predict how life may exist on other planets? A paper in Interface Focus takes questions like this out of science fiction and into scientific laws. santafe.edu/news-center/... 67323
Reposted by Manuel BaltieriDavid Pfau @davidpfau.com · 23/12/2024So, they set up the ARC prize specifically to have limits on how much compute you can use, required the submissions to be open-sourced, and then OpenAI just went ahead and crushed it with >$1m of compute from a closed model with no idea how it works anyway? Do I have that right? 7855
Reposted by Manuel BaltieriPhil Lewis @phillewis.bsky.social · 20/12/2024For years, former Black Panthers have feared that the FBI is still keeping tabs on them decades after its notorious COINTELPRO program. Malik Rahim had never been able to prove he was targeted. Now, government docs show he was correct.thenation.comWhen the Feds Are Still WatchingFormer Black Panthers fear the FBI is still keeping tabs on them decades after COINTELPRO. In at least one case, they were right. 16497293475